User and Customer Needs
2.d Satisfies Required Analysis Level
1.a Quick Feedback
1.d Easy to Learn
1.f Feature Specific Feedback
1.g Complex Data Inputs
1.e Integrated DFM Intuition
1.b Intuitive UI / UX
1.c CAD Integration
1. h Ability to trust the insights
2.b Integration with existing software
2.c Reduction of Time is Clear
2.a Reduction of Cost is clear
Near-instant feedback
Features:
Considerations:
Integrated CAD plugin
Confidence score
Back-end analysis
1a, 1c
1b, 1d, 2b
1a, 1h
1e, 1f, 1g, 2a, 2c, 2d
From interviews with users and customers, we were able to generate a list of crucial considerations for the software, leading to a list of key features.
Users can interact with effio directly or through our custom CAD plugin that simplifies the process, causing minimal disruption.
User Journey - How to Use effio
Benchmarking our software against the main competing software
By designing a sample part with predetermined errors, we wanted to see how well the average mechanical engineer and CNC operator would perform at identifying opportunities for improvement. The table below shows the average results from that testing, and should not be unsurprising given that DFM education for mechanical engineers is very poor, and many manufacturers are less concerned with finding issues to fix, and more concerned with finding ways to machine whatever they’ve been given.
Competitors and Differentiation
Engineer designing part in CAD
Significant changes to part?
Y
Y
N
N
Plugin input
Plugin reminder
effio drag & drop
effio advanced inputs
effio summary
effio feature specific analysis
Make changes
Concerning feedback?
Plugin summary
Considerations: 1a, 1c, 1h, 2a, 2c
Considerations: 1c, 1d, 2b
Plugin Reminder
The plugin has a traffic light system that prompts the user to audit their part when a large change in geometry is detected.
= unvisited
= current
= visited
Plugin Summary
Once the analysis is complete, the engineer will see a summary of the audit in their CAD, helping them spot any new issues.
Considerations: 1c, 1g, 2b
Plugin Input
The user can input some simple parameters such as material, manufacturing process, and quantity.
Considerations: 1f, 1h, 2a, 2c, 2d
effio Summary
In effio, the engineer can see a more comprehensive summary of the part, including key measurements.
effio Feature-Specific Analysis
Finally, the engineer can see feature-specific feedback, with suggestions for mitigation and a measure of importance.
effio Advanced Inputs
effio supports advanced inputs that enable more bespoke, advanced, and useful feedback for the engineer.
Considerations: 1b, 1d
Considerations: 1b, 1d, 1g
effio Drag & Drop
For convenience, an engineer can import a new file by simply dragging and dropping from a file/desktop right into effio.
For engineers not using the CAD plugin:
Considerations: 1f, 1h, 2a, 2c, 2d
DFM Pro
Sybridge
Protolabs
aPriori
effio
External fillets
Hole diameter too small for given material
Faces tapered in z axis
Requires soft jaws
Sharp internal corners
Part height close to stock dimension
Size requires too large machine vice
Off-axis feature
Pocket depth to fillet radius ratio is high
Undercut Feature
Thin features subject to vibration
Inconsistent pocket fillet radii
Single feature requires multiple DoF
How easy to measure
14
7
CNC Operator
Competing Software
4
Mechanical Engineer
4
effio
17 key assumptions were made, and the most assumptions are included here:
Assume arbitrary computations are reasonable proxy for geometric analysis
Assume highest available market price for chosen GPU
Assume no batching of requests
Assume part specs equivalent to largest in dataset
Assume no quantisation, sparsification, or distillation of model
Assume no compression algorithm in use
Assume use of excessively powerful CPU
Assume no concurrency or parallelisation of deterministic geometric analysis
Price per Analysis
Impact
Uncertainty
Total Price & Discussion:
The cost estimate for is much lower than the LLM proxy used in the finances (although this price still used for conservativism).
The geometric analysis is lightweight enough to be run locally, although for large parts it is advised to use a web-hosted service.
1.
2.
3.
4.
6.
8.
7.
5.
Part Data Experiment
Used a dataset of 10,000 STEPs to understand the compute time:
Wrote a script to loop over the first 1,000 parts and perform arbitrary computations over every edge and face
For each part we keep track of the number of faces and edges, time taken, and compute resources required (CPU utilisation and RAM)
Average Compute:
RAM (LPDDR4X 4266 MT/s): 245 MB
CPU (M1 3.2GHz): 98% of 1 Core
Part Statistics:
Time taken (secs): 0.0469 (av) 3.3955 (max)
No. of faces: 90.79 (av) 5,885 (max)
No. of edges: 477.97 (av) 32,732 (max)
Machine Learning Pricing
Chosen GPU: Nvidia A40000 - 19.2 TFLOPS (conservative), 16GB DDR6 memory
Cost of running model on single (large) part: $0.00609
Geometric Analysis Pricing:
Chosen VM: 4GHz CPU & 1024MB RAM for $0.0000165 per second
Using highest time from data experiment (3.3955 seconds) -> $0.000056
Figure 1: part data experiment
£0.0049
[Cost of a one geometric analysis and machine learning analysis]
The analysis has been split between deterministic geometric checks and machine learning, modelling the machine learning on work done by Autodesk Research, called the UV Net. While this isn’t the approach the team are using, it serves as a conservative and largely representative proxy for the structure of our own.
Analysis Process
Each time a user clicks “run analysis”:
1. (If from CAD) local conversion to STEP format, part ported to effio
2. Local generation of model inputs e.g. UV maps in effio, data ported to server
3. Model weights loaded from memory on virtual machine
4. Forward pass for each face and edge using generated model inputs
6. Deterministic geometric checks on STEP file
7. Results of ML and geometric analysis combined
8. Necessary feedback fetched from database and sent to user’s machine
9. Results of analysis are displayed locally on user’s instance of effio
NB. Exact order of analysis will depend on the process the part is being analysed for
This section shows more detailed and robust calculations for the COGS of our service (and benchmarking the computational complexity), yielding a more favourable result than the LLM proxy used in our financial projections.
Market sizing
£532 Billion
TAM
SAM
SOM
£2.18 Billion
£8.72 Million
Total Addressable Market (TAM)
European share of the total Injection Moulding1 4 , CNC Machining2 5 and Die Casting3 6 Markets, taking into account the growth rate7 8 9
Serviceable Available Market (SAM)
UK specific market share10 11 multiplied by the percent of total costs committed to at Design Freeze12 multiplied by the percent in cost reduction13
Serviceable Obtainable Market (SOM)
Percent of hardware centred companies in London & Cambridge compared to rest of UK14 multiplied by the market capture rate15
Figure 1: Market Sizing diagram
https://www.fortunebusinessinsights.com/injection-molded-plastics-market-101970
https://www.fortunebusinessinsights.com/industry-reports/computer-numerical-controls-cnc-machine-tools-market-101707
https://www.imarcgroup.com/die-casting-market
https://www.grandviewresearch.com/horizon/outlook/injection-molding-machine-market/europe#:~:text=In%20terms%20of%20revenue%2C%20Europe,molding%20machine%20market%20in%202023.
https://www.grandviewresearch.com/industry-analysis/europe-computer-numerical-control-cnc-machine-market
https://www.grandviewresearch.com/industry-analysis/europe-aluminum-die-casting-market-report#:~:text=The%20Europe%20aluminum%20die%20casting,5.1%25%20from%202024%20to%203030.
https://www.grandviewresearch.com/horizon/outlook/aluminum-die-casting-market/united-states#:~:text=In%20terms%20of%20revenue%2C%20U.S.,terms%20of%20revenue%20in%202030.
https://www.grandviewresearch.com/horizon/outlook/cnc-machining-centers-and-turning-centers-market/united-states#:~:text=The%20U.S.%20cnc%20machining%20centers,5%25%20from%202024%20to%203030.
https://www.grandviewresearch.com/industry-analysis/north-america-plastic-injection-molding-machine-market-report
UK manufacturing sector climbs to eighth in world - Plant & Works Engineering
European Union Manufacturing Output 1991-2025 | MacroTrends
percent of total costs committed to at Design Freeze
5% cost reduction (conservative)
100 Top Hardware Companies in United Kingdom · March 2025 | F6S
1% market capture (conservative)
Features and Differentiation
Contents